If you’ve been in the self-hosting community for a while, you may already be aware that you can run firewalls, DNS servers, and traffic monitoring tools on local hardware to add an extra layer of security to your home network. But your network setup and online privacy aren’t the only things you can safeguard with self-hosted services.

With their monitoring, tracking, and notification-based alert facilities, surveillance platforms help you keep a vigilant eye on your home. Plus, you can self-host these tools to add an extra layer of privacy to your living space. If you’re looking for a self-hosted surveillance solution, you won’t be disappointed with Frigate.

4 It's a lightweight container

Even a Raspberry Pi can handle this NVR tool

Home servers built from old enterprise-grade hardware are ideal for hardcore experiments, but not everyone has the space or money to run power-hungry components 24/7. If you’re using an ARM-based NAS or an SBC as your primary self-hosting workstation, Frigate’s lightweight nature can deliver decent performance in your NVR workloads. For a budget-friendly security system, you can pair one of the newer Raspberry Pi models with an affordable IP camera, while the resource-efficient Frigate container can handle the software side of your surveillance solution.

3 Solid detection and tracking facilities

Gotta keep false positives to a minimum

Although the alert mechanisms built into surveillance platforms are a lifesaver, it’s also easy for your self-hosted security tool to mistakenly send you notifications about intruders. Sure, Frigate isn’t free from the curse of false positives, but it provides some extra means to curb mistaken reports. For instance, Frigate lets you configure filters, masks, and zones to increase the accuracy of its reports. And that’s before you include its solid auto-tracking provisions that can move PTZ-compatible cameras to follow moving objects.

2 Clean, uncluttered UI

That’s also easy to configure for Docker maestros

If you’ve used ZoneMinder in the past, you may have realized that it’s also armed with decent tracking and object detection facilities. But if you’re not into hardcore self-hosting workloads, it’s easy to get intimidated by its user interface. In contrast, Frigate has a neat and uncluttered web UI that, despite being more beginner-friendly, lets you configure most of the essential settings. The only caveat is that you’ll have to modify the config.yml file to add more cameras instead of reying on menus inside the web UI, though you shouldn't have too much trouble if you’ve ever tinkered with Docker Compose or Kubernetes config files.

1 Compatible with AI accelerators

At last, a decent use-case for AI!

The ability to interface a wide range of AI-accelerated tracking provisions into my surveillance setup was single-handedly the reason I swapped ZoneMinder for Frigate. By default, it offers OpenVINO and Tensor-powered AI acceleration for CPUs and GPUs (respectively), though it's the support for the Hailo-8L chip and Coral Edge TPU that stands out the most. If the name sounds familiar, Hailo-8L is the NPU powering the official Raspberry Pi AI Kit. While pairing the AI module with Frigate is a little harder than it looks, it’s hands-down the best way to put your spare AI module to good use. Alternatively, you can go for the simpler Coral Edge TPU module, which is also compatible with mini-PCs and NAS units.

Which NVR tool is your favorite?

Although I’ve largely moved over to Frigate, I’ll admit that I still have a soft spot for MotionEye and ZoneMinder. The former combines the best aspects of ZoneMinder and Frigate’s UI and can be set up as a full-fledged OS on older Raspberry Pi models. On the other hand, ZoneMinder is a powerful utility when you want to manage multiple cameras in your security system, though you’ll need a beefy system to use its advanced NVR features. However, Frigate’s lightweight nature, robust tracking, and support for AI accelerators make it perfect for my surveillance needs. Just remember to limit the storage capacity of the recordings when creating the Docker container, or you might end up choking your NVR hub with hours of footage.